No Script, No Fear, All Opinion
• ## EEVblog #102 – DIY Constant Current Dummy Load for Power Supply and Battery Testing

Posted on August 1st, 2010

Dave grabs a few junkbox parts and builds a useful constant current load for switch mode power supply, battery testing, and other applications.

Be Sociable, Share!

### 54 responses to “EEVblog #102 – DIY Constant Current Dummy Load for Power Supply and Battery Testing”

• Dave another excellent episode. On the audio segment you said you were possibly planning a return to hobbyist kits. I believe this with some of the digital assisted aspects your mentioned would be a wonderful for kit as bench top tools for a hobbyist. While watching I keep wondering if adding in a parallel variable capacitor or some induction would be a good way to show the effects of parasitic effects on power systems.

• I made something simular a while back, only it is a bit bigger :p. It go’s to 16A.

• Hi Dave and all,

I’m kind of new to electronics and just an example with your circuit, if we wanted to make the heat more uniform instead of it being all on the transistor, would hooking up a heap of diodes (rated 1A) and 2 transistors (rated 1W Tambient) work as the picture below?

http://img72.imageshack.us/img72/3967/diodes.png

Let me know what could go wrong here?

Alex.

• Alex, there are two problems with your circuit:

1. Constant current sources are supposed to keep their output current constant regardless of the output voltage. Series diodes will limit the minimum operating voltage of your circuit.

2. The total amount of power that is dissipated in the circuit is fixed and it does not depend on the components used. All this power is converted into heat and it is the role of the heat sink to push this heat out of the system so that the components stay within their specified operating temperature range. The only thing your circuit achieves is helping the heat sink to stay cooler at the expense of potentially overheating poorly cooled diodes. The correct solution would be to use a larger heat sink, forced air flow or a larger transistor. Look at the Stynus’ website for an example.

I like the way you’re exploring the problem space and asking questions. Both are very important when learning about electronics.

• Just use another MOSFET + heat sink in parallel.

Adding the diodes will force you to up the input voltage to compensate for the _fixed_ voltage drop across them. With that many diodes you couldn’t use it with a 5V power supply.

• I’m a bit surprised you got away without a resistor between the opamp and MOSFET gate to prevent oscillations. The LM324 is only good to drive 50pF and I would have thought the gate capacitance was higher than than. But if it works …

• Hi Dave,
I used to build stepper motor drivers when I was a kid, and one day I made one using this circuit as a constant current sink, but it oscillated. I wonder why yours doesn’t oscillate, maybe it is due to different components or a different external circuit.

Eventually I figured out that the reason why mine oscillated was that there is a pole in the opamp due to the internal compensation capacitor, and there is another pole due to the RC circuit formed by the output resistance of the opamp and the gate capacitance of the MOSFET. Each of these two poles gives close to 90 degrees of phase shift at high frequencies, and there are bound to be some other less significant poles due to parasitics which will give a few more degrees so that the total phase shift around the loop due to the poles adds up to 180 degrees at some high frequency, and 180 degrees of phase shift turns what should be negative feedback into positive feedback, and oscillation.

The nicest way of solving this would be to get rid of the pole of the op-amp compensation by using a transconductance amplifier instead, but chip companies for some reason don’t market these much as products (although chip designers use them all the time inside bigger chips, about as often as opamps). The only ones I could find available to buy were CA3080 (obsolete) and LM13700, both of these need dual supplies. It would be possible to build one with a few transistors but the component count would be a bit high then.

It’s good that your one doesn’t oscillate, maybe your FET has less capacitance than the ones I was using, leading to less phase shift. The load can have an effect too, if the drain of the MOSFET is connected to a high impedance load (like the coil of a stepper motor) then the gain from the gate to the drain of the MOSFET is high, and the Miller effect then makes it look like the gate capacitance is even higher.

Chris

• Dave you should make and sell some kits with helpful electronics testing tools just like this one.

• …Where can I get that awesome DaveCAD program?!?

• webkraller,

If you have a compatible system of paper and marker, I have a free serial key for DaveCAD!

• Karl (not that Karl, the other Karl)

I am actually using an LM317 for a 1A current source / load.

You don’t need a power wire-wound resistor. The same trick you did with the 1 Ohm resistor works, paralleling cheap 1% resistors. I use a bunch of switches to parallel more or fewer resistors to adjust the current.

• Another really cool video, Dave. Quick question, what would be different if you used (or didn’t use) thermal compound between the heatsink and the MOSFET? Would it be very slightly cooler with thermal compound?

Thanks.
~John

• what would be different if you used (or didn

• Whoops, no. The 11 W heat load being dumped into the heat sink hasn’t changed, and ambient temperature hasn’t changed, so if we discount direct heat dissipation from the MOSFET by other paths then the heatsink must remain at the same temperature. The entire effect of a poor thermal path between the MOSFET and the heatsink is that the MOSFET will get hotter.

I like the idea of this project and I am thinking of making one. However, for testing batteries I will aim to make the load resistance lower than 1 ohm so it can sink reasonable currents from lower voltage cells like NiMH.

• I like the DavCAD also. Need to get that ASAFP.

Hey!! I’d like to swap my junk box for yours. All my junk is junk! You’ve got neat junk in yours. How ’bout it!?!?!?

I was just reading up on Op Amps today. Perfect timing for your blogs topic.

• David Robertson

(This is me re-posting an earlier comment i made here which was lost (i assume) as Dave transferred the site)

• Michael Thompson

I think I posted one that got lost as well.

How well would this project scale up to handle 10s or 100s of amps?

A buddy is giving me what he calls a “big old industrial power supply” that I’d like to use for my ham radio bench setup, but if it’s that old i think it needs to be looked over very carefully.

This really is a cool project!

• Just found your blog. AWESOME! The kind of circuit walk through is totally invaluable for a hobbyist.

Well done.

• Very nice job. I’ve built and used a dummy load, with just a 10K pot and a 2N3055. Must build one more complete like yours one of these days :).

• Hello guys-Dave. Found that blog a couple of weeks ago. Nice work Dave, BTW your accent reminds me of my cousin in Australia.

Guys i’m working on a dummy load myself. I have decided to make it with high power resistors. It will have a maximum power draw of 150 watt. The think that bothers me is if i can use PWM to drive the source current to the load and if yes what will be a good frequency

• Hi there,

Dave I like your hands-on approach – goes directly to my amteurish brain.
So I just built one of these and it is oscillating (I can see spikes on the +5 rail) and the output from the lm324 is oscillating somewhere between zero V and 1 V. A Filter cap between + and – din’t help:(

Any Ideas Anyone?

• The LM324 doesn’t like driving capacitive loads much. Try a series resistor on the opamp output.
Or a brute force 10n cap between the inverting and noninverting inputs might stop it too.
Dave.

• As always, a great blog, Dave.

I followed your heat sink example, but I do not understand why a temperature rise is important.

So we have a wattage being dissipated by the FET and that results in a specific temperature increase, and the heat sink conducts the heat to the air.

Why do we care about the temperature? (I’m sure there’s a reason — please enlighten us.)

Thanks for another great blog!

Dave C.

• The operating life of most components decreases drastically with an increase in temperature. So if your FET is running at a die temperature of say 125degC, then it’s not going to be as reliable as one that runs at a lower temp like 80degC
As far as the system design inside the box goes, other components can also be affected by nearby heat sources. Electrolytic capacitors is a classic example of this.
So it’s desirable from many aspects to keep your designs at as low a temperature as possible.

Dave.

• Hi Dave,

I’m a hobbyist/DIY’er with primitive electronics knowledge going all the way back to the eighties and never really developed since. Lately I’ve been trying to crash course myself up to speed. I’ve come across your blog and it is quite interesting , so thank you!!!!

Re: the Dummy Load for Power Supply – How would you set up testing a LCD HDTV SMPS seperate from the TV and simulate a load from SMPS that transforms AC source of 120V~240V AC +/- 10% @ 50/60 HZ into DC 5V & 12V& 24V source.

• Hi Dave,
First of all, thanks for excellent blog and excellent videos. I love your latest blog on switchmode DC-DC power supplies. It is very timely and appreciated information for my latest projects.

Are you selling any of the blank circuit boards you show in this video?

I have been wanting to build a battery performance logger, and this looks like a pretty nice start.

Thanks,

• No plans to sell that board, sorry.

• Doesnt seem to be working on my IE7. Just thrashes incessantly, refreshing the address bar non-stop. View source shows about 6 lines of curious JavaScript. I can be from may laptop but the page its ok on Mozila, i cant undestend someting im on alexa now and your rank is verry big, i found you blog on second page of google .Andrei from Italy

• Hi Dave,

On the similar lines, I think, we can make a programable power supply using a micro-controller. The output from Micro to DAC to Op-amp to transistor. The feedback is measured by ADC to maintain a regulated voltage output. The voltage output can be set by key pad or through commands from serial port!!

• Nice video, I just made one capable of 500 watts, ~40 amps @ 12v for testing computer Power supplies. I needed a ‘slightly’ more complicated cooling solution though…
Check it out:
http://img843.imageshack.us/g/img0733um.jpg/

• Thanks for sharing your thoughts in that last post. You have a talent for making a hard subject clear to others. I enjoy reading the posts from a guy who has the same flair for explaining things.

• Hi Dave!

Thanks for the idea, just made one myself too. I modified your lm358 based schematic, added current meter on ICL7107, and i must say it works perfect Range 10mA to 20A, 20A is just a matter of cooling and number of mosfets.

Here is the link if you don’t mind, no english description yet http://diy.elektroda.eu/sztuczne-obciazenie-o-regulowanym-pradzie/

• Something to watch out for with this kind of circuit. At the moment the load is being connected, there is no voltage yet over the transistor and the gate of the fet is fully open. This will generate a current peak before the opamp has time to squeeze the gate.

• I’ve watched the video several times and event breadboarded-up the circuit but I’m still unclear about what exactly the MOSFET and the 2nd op-amp are doing.

I understand the the op-amp “wants” it’s inputs to be the same, so if there’s 1V on the “+” input it’s going to do “something” to the output until it sees “1V” on the “-” input.

But what is actually happening here? What is the role of the MOSFET? Why if there’s a 12V load attached to the drain is 12V not seen on the source as well?

Thanks.

• Hey Dave,

Thanks for the idea. I just built up my own variable current load and through it in an Altoids box.

http://embeddederic.blogspot.com/2011/07/everything-including-kitchen-current.html

• What should or could I do to use this with a PSU arround 50V. The problem I think ‘would arrise is that the Vgs would we out of spec.

Must I go to a P-channel mosfet? But that wouldn’t work with the opamp.

Is there a way arround this?

• Great write-up, I’m normal visitor of one’s blog, maintain up the excellent operate, and It’s going to be a regular visitor for a long time.

• Is it good idea to use this as first phase (constant current) for SLA battery charger? It would be controlled by uC, than second phase would be constant voltage.

• Dave love your videos… I’ve been playing around with a similar circuit (smaller current) only I used the second op-amp to amplify the feedback from the sense resistor to be closer to 5v so my PWM output would be utilizing closer to it’s full range… am I fooling myself into thinking this is a good idea? does the amplification of the sense voltage nullify any gain I think I’m getting with the full range on my PWM?

Always anxious to see your next video

• I just scaled this up on paper to go up to 9 amps. I found the real limiting factor is the sense resistor. It will have to dissipate I^2*R watts, so for 5 amps with a 1 ohm resistor you’re at 25 watts. I’m going to try paralleling 40 (yes forty) 1/2 watt 10 ohm resistors to give me a 0.25 ohm 20 watt resistor. It will reach 20 watts around 8.95 amps.

• Yeah, you can get to serious power real quick. Big heatsinks in order for both resistor and FET!

• Hi Dave.

With consideration of the polarity can you put such a circuit in series with a PSU as a current limiter? I have a 1.5v – 30V 20A power supply that I want to tame for more delicate work with an upper limit of 1A? Putting this circuit in series when I need it sound like a good way of taming the 20A limit to 1A, fully variable.

I later intend on modifing my PSU to have a 20A non variable, and then be able to switch in the 1A variable limiter.

Nick

As you hoped, it didn’t work! Then I learnt something about MOSFET power handling, and now it does.

• Awesome!

• Hi
Can be MOS FET transistor (MFT3055VL) replaced with another ?
If yes, what characteristics ?
Thank you and all the best

MikeM

• Could you elaborate a little more about how the micro would manage constant power? Wouldn’t you need a measurement of the voltage at the mosfet drain as well? Then you could use the known current and the device output voltage to calculate power, right? Thanks!

• Well, followed DaveCad for my own design, initially had 100mA total load current, simulated in LTSpice to see where I was going wrong, realised that my mistake, 1st DaveCad copy was correct, circuit LTSpice layout was incorrect as I was again getting about 100mA simulated, spotted my mistake and corrected, then looked on the physical circuit and fount the SAME mistake, Go figure!!!!

• Hi,

What would be a better op-amp for this? And mosfet?

• Hi all,

I am also thinking of building this to draw a current from a DC motor whose armature is spun by a separate motor (this is to simulate a load acting on a generator). In such a case what are the factors should I be concerned about?

How does the constant load respond to a varying voltage (like one produced by a mechanically spun DC motor) at the drain of the MOSFET?

Best regards

• Hi, I’m thinking of building a dummy load of about 1kW for 48V battery testing. does similar design feasible?

• Hi, can this dummy test negative rails or it is necesary to change the fet? Thanks.